As Paul Krugman likes to point out, conservative leaders have a bad habit of just making things up: global warming isn't happening, tax cuts pay for themselves, quantitative easing will lead to hyper-inflation etc.. David Leonhardt tells readers that at the Upshot section of the NYT, which he edits, they are committed to calling both sides out when the facts don't support their claims.

And he is taking the occasion to beat up on liberals on the relationship between marriage and happiness. He proudly calls attention to a new study that purports to show that married couples are happier on average than people who are not married, and this is even after controlling for states of happiness before they were married.

It's not clear exactly what liberal view Leonhardt thinks he is challenging. There is an obvious survivor bias in a long marriage. We expect that people in unhappy marriages are less likely to stay married, so this study has effectively found that people in happy stable relationships are happier on average than people not in happy stable relationships. Are there liberals who feel it is important to argue that this is not true?

There is a problem that liberals, like any believer in logic, may have with policy prescriptions that could be mistakenly based on this finding. For example, it certainly does not follow, based on this research, that the government should make it more difficult for couples to divorce. The point is that happy couples are happier, if we forced unhappy couples to remain married, it does not follow that they would be happier than if they were separated.

We may also think that the government should foster happy couples by having subsidies for marriage. But this effectively amounts to penalizing the people who are already unhappy because they are single. After all, someone has to pay for these subsidies, which means that on average we would have money flowing from already unhappy single people to our happy couples. Is this good policy?

It's also not clear that there is importance to marriage as opposed to a stable relationship. In the U.K. (where the subjects for the study lived), like the U.S., most people in stable relationships tend to get married. However in other countries this is not a cultural norm. Are we supposed to believe the sheet of paper makes people happy? Did the Wizard of Oz make the straw man smart when he handed him a diploma?


As one of his other examples of liberals holding views that defy the evidence Leonhardt goes after my friend Richard Rothstein for purportedly arguing "education is overrated," while still sending their kids to expensive colleges. This one really misses the mark also.

In the piece to which Leonhardt links, Rothstein argues that getting more college grads is not the answer to the country's inequality problem. (If Leonhardt thinks Rothstein argued that college grads don't on average have higher incomes than non-college grads, he should re-read the piece.) Rothstein's point was that even college grads had been seeing stagnating wages up to that point in the business cycle (2008). That pattern has continued through 2013, the most recent year for which annual data are available. While per capita income has risen by more than 11.0 percent since 2000, none of this has shown up in the paycheck of the typical college grad. The gains have instead gone to profits or more highly paid workers, like CEOs and Wall Street types.

It is ironic that Leonhardt would make this complaint on a day when the Bureau of Labor Statistics released data showing that the employment to population ratio for college grads is down by 0.2 percentage points over the last year. By contrast, it is up by 0.5 percentage points for high school grads and by 2.1 percentage points for workers without a high school degree.

So if the question is whether a young person should try to get a college degree, I suspect Richard, like most of the liberal-lefty types I know, would unambiguously answer yes. If the question is whether more people getting college degrees is the answer to inequality, we would say no, based on the data.